Quantitative Comparison of Two Cortical Surface Extraction Methods Using MRI Phantoms Simon F. Eskildsen and Lasse R. Østergaard Dept. of Health Science and Technology, Aalborg University, Denmark Abstract. In the last decade several methods for extracting the hu- man cerebral cortex from magnetic resonance images have been pro- posed. Studies comparing these methods have been few. In this study we compare a recent cortical extraction method with FreeSurfer, which has been widespread in the scientific community during recent years. The comparison is performed using realistic phantoms generated from sur- faces extracted from original brain scans. The geometrical accuracy of the reconstructed surfaces is compared to the surfaces extracted from the original scan. We found that our method is comparable with FreeSurfer in terms of accuracy, and in some cases it performs better. In terms of speed our method is more than 25 times faster. 1 Introduction Reconstruction of the human cerebral cortex from magnetic resonance (MR) im- ages facilitates morphometric studies and brain mapping, and provides intuitive visualisation of the human brain for the use in e.g. surgical planning. Since the nineties a number of algorithms has been developed for extracting the bound- aries of the cortex from MR images [1,2,3,4,5,6,7]. FreeSurfer has been around for more than seven years, and has, due to the fact that it is freely available, become widespread in the scientific community. We have recently published a method (henceforth designated Fast Accurate Cortex Extraction (FACE)), which resembles FreeSurfer in many aspects, but is significantly improved in terms of computational speed [8,9]. When performing morphometric studies the accuracy of the cortex recon- structions is very important. Therefore, it is of interest to investigate how well ACE performs in terms of accuracy compared to FreeSurfer. Quantification of the accuracy is difficult as the ground truth is rarely available. A means to mea- sure the accuracy is using phantoms resembling real neuroanatomical data. Lee et al. [10] compared FreeSurfer [4], CLASP [7] and BrainVISA [2] using gener- ated phantoms. They found that CLASP was more accurate than BrainVISA and FreeSurfer. However, CLASP is not publicly available, while the two other methods are. FreeSurfer performed second best in the study. In this study we compare our method, ACE, to FreeSurfer using realistic phantoms generated from real MR scans. N. Ayache, S. Ourselin, A. Maeder (Eds.): MICCAI 2007, Part I, LNCS 4791, pp. 409–416, 2007. c  Springer-Verlag Berlin Heidelberg 2007